CN109189944A - Personalized recommending scenery spot method and system based on user's positive and negative feedback portrait coding - Google Patents
Personalized recommending scenery spot method and system based on user's positive and negative feedback portrait coding Download PDFInfo
- Publication number
- CN109189944A CN109189944A CN201811131075.8A CN201811131075A CN109189944A CN 109189944 A CN109189944 A CN 109189944A CN 201811131075 A CN201811131075 A CN 201811131075A CN 109189944 A CN109189944 A CN 109189944A
- Authority
- CN
- China
- Prior art keywords
- sight spot
- user
- spot
- attribute
- positive
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The present invention proposes a kind of personalized recommending scenery spot method based on user's positive and negative feedback portrait coding, includes: acquisition user to the history evaluation information at sight spot and the Tourism Attribute data at sight spot；Obtain the positive and negative evaluation sight spot of the user；Triple is converted by all sight spots and its Tourism Attribute value to construct sight spot knowledge mapping；Sight spot entity, attribute and attribute value in triple is trained by score function, sight spot entity, attribute and attribute value is made to be converted to vector representation；User is obtained to the positive and negative feedback portrait coding at sight spot；Similarity calculation, which is done, with sight spot entity vector using positive feedback portrait coding obtains the sight spot that user likes；Negative-feedback portrait coding is recycled to optimize the sequence.The present invention encodes the negative-feedback of sight spot feature portrait by user, and increases on the basis of favorite interest list sight spot similarity calculation that user dislikes and optimization obtains consequently recommended list, provides more accurate recommending scenery spot for user.
Description
Technical field
The present invention relates to the technical fields such as machine learning, knowledge mapping and intelligent recommendation, are specifically related to one kind and are based on
The personalized recommending scenery spot method and system of user's positive and negative feedback portrait coding.
Background technique
With people's living standard and living-pattern preservation, more and more people need outdoor activity, such as watch movie, have dinner
Movable with tourism etc., people become increasing for the personalized recommendation demand of location information, can recommend when people go on a journey
Suitable sight spot is particularly important.Therefore, how user constructed according to the evaluation feedback information for having accessed sight spot of user
Portrait, this is the key that personalized recommending scenery spot system.Traditional location-based recommendation, most of is all according to sight spot temperature
Statistical information is recommended, and is not well positioned to meet the requirement of user individual, while also not accounting for recommending sight spot itself
Characteristic cannot accurately give user's recommendation information.
Summary of the invention
In view of the foregoing deficiencies of prior art, the purpose of the present invention is to provide one kind is drawn based on user's positive and negative feedback
As the personalized recommending scenery spot method and system of coding, to solve in the prior art based on recommending scenery spot essence in the recommended method of position
Degree is not high to experience the problems such as degree is low with recommendation results personalization and sight spot.
In order to achieve the above objects and other related objects, the present invention provide it is a kind of based on user's positive and negative feedback portrait coding
Personalized recommending scenery spot method, method includes the following steps:
User is acquired to the history evaluation information at sight spot and the Tourism Attribute data at sight spot, and is pre-processed；Further according to
Every user obtains the positive and negative evaluation sight spot of the user to the height of the figure of merit at sight spot；All sight spots and its tourism are belonged to
Property value is converted into triple to construct sight spot knowledge mapping；
The triple in the knowledge mapping of sight spot is mapped in characteristic vector space using the method that network representation learns, it will
Sight spot entity, attribute and attribute value in triple are trained by score function, turn sight spot entity, attribute and attribute value
It is melted into vector representation；
According to user to the height of the figure of merit at sight spot, the corresponding weight coefficient of sight spot entity vector is set, then
Sight spot entity vector is calculated with corresponding weight coefficient, user is obtained respectively and the positive and negative feedback portrait at sight spot is compiled
Code；
Similarity calculation, which is done, with sight spot entity vector using positive feedback portrait coding obtains the sight spot that user likes, and according to
Similarity is from high to low ranked up sight spot；It recycles negative-feedback portrait coding to optimize the sequence, is finally arranged
Sequence.
Preferably, the acquisition user is to the history evaluation information at sight spot and the Tourism Attribute data at sight spot, and carries out pre-
Processing；Further according to every user to the height of the figure of merit at sight spot, the positive and negative evaluation sight spot of the user is obtained, is specifically included:
User is acquired to the history evaluation information at sight spot and the Tourism Attribute data at sight spot；
For the attribute value data at user, sight spot and sight spot, ID is set；
Sight spot is divided into according to history evaluation information of the user to sight spot and likes sight spot and disagreeable sight spot.
Preferably, described to convert triple for all sight spots and its Tourism Attribute value to construct sight spot knowledge mapping, have
Body includes:
The form that the Tourism Attribute data triple (P, V, Q) at sight spot and sight spot will be collected indicates that wherein P is sight spot
Entity, V are attributes, and Q is sight spot attribute value；The sight spot entity and sight spot attribute value table of triple (P, V, Q) are shown as node, relationship
It is expressed as side, two nodes are connected using side, constitutes a network about sight spot, the i.e. knowledge mapping at sight spot.
Preferably, it is described using network representation learn method by the triple in the knowledge mapping of sight spot be mapped to feature to
In quantity space, sight spot entity, attribute and the attribute value in triple are trained by score function, make sight spot entity, belongs to
Property and attribute value are converted to vector representation, specifically include:
Scenic spot entity, attribute and attribute value in triple is converted into digital shape word；
The ID of sight spot entity, attribute and attribute value in taking-up triple；
Sight spot entity ID, Property ID and attribute value ID are indicated with a low-dimensional real-valued vectors to obtain sight spot entity, belong to
The vector of property and attribute value, is normalized institute's directed quantity；
According to the distance properties of triple construct scoring function, training triple, with update the sight spot entity, attribute and
The vector of attribute value indicates.
Preferably, it is described according to user to the height of the figure of merit at sight spot, the corresponding power of sight spot entity vector is set
Weight coefficient, then calculates sight spot entity vector with corresponding weight coefficient, obtains user respectively to the positive and negative of sight spot
Feedback portrait coding, specifically includes:
The sight spot entity vector of positive and negative evaluation is obtained respectively；
The weight coefficient of the scape entity vector of positive and negative evaluation is respectively set；
Positive and negative portrait is obtained according to the sight spot entity vector of positive and negative evaluation and respective weights system-computed to encode.
Preferably, described to do similarity calculation using positive feedback portrait coding and sight spot entity vector and obtain what user liked
Sight spot, and sight spot is ranked up from high to low according to similarity；Negative-feedback portrait coding is recycled to optimize the sequence,
It is finally sorted, is specifically included:
It draws a portrait to encode to calculate using positive feedback and obtains the alternative recommendation sight spot of user；
It is drawn a portrait using negative-feedback and encodes the sight spot that calculating acquisition user does not like；
Using the alternative recommendation sight spot of negative-feedback portrait code optimization, obtains and recommend sight spot.
In order to achieve the above objects and other related objects, the present invention also provides one kind based on user's positive and negative feedback portrait coding
Personalized recommending scenery spot system, which includes:
Data acquisition and procession unit, for acquiring user to the history evaluation information at sight spot and the Tourism Attribute number at sight spot
According to, and pre-processed；Further according to every user to the height of the figure of merit at sight spot, the positive and negative evaluation scape of the user is obtained
Point；Converting unit, for converting triple for all sight spots and its Tourism Attribute value to construct sight spot knowledge mapping；
Triple in the knowledge mapping of sight spot is mapped to feature by construction unit, the method for being learnt by network representation
In vector space, sight spot entity, attribute and the attribute value in triple are trained by score function, make sight spot entity,
Attribute and attribute value are converted to vector representation；
Sight spot entity is arranged for the height according to user to the figure of merit at sight spot in positive and negative feedback portrait coding unit
Then sight spot entity vector is calculated with corresponding weight coefficient, is used respectively by the corresponding weight coefficient of vector
It draws a portrait and encodes to the positive and negative feedback at sight spot in family；
Recommendation unit obtains user for sight spot entity vector doing similarity calculation using positive feedback portrait coding and likes
Sight spot, and sight spot is ranked up from high to low according to similarity；Negative-feedback portrait coding is recycled to carry out the sequence excellent
Change, is finally sorted.
Preferably, the data acquisition and procession unit includes:
Data acquisition unit, for acquiring user to the history evaluation information at sight spot and the Tourism Attribute data at sight spot；
Configuration unit, for ID to be arranged for the attribute value data at user, sight spot and sight spot；
Division unit likes sight spot and disagreeable scape for being divided into sight spot according to history evaluation information of the user to sight spot
Point.
Preferably, the converting unit includes:
First conversion unit, the shape of the Tourism Attribute data triple (P, V, Q) for sight spot and sight spot will to be collected
Formula indicates that wherein P is sight spot entity, and V is attribute, and Q is sight spot attribute value；
Knowledge mapping acquiring unit constitutes a network about sight spot, i.e., for two nodes of triple to be connected
The knowledge mapping at sight spot, wherein the sight spot entity and sight spot attribute value table of triple (P, V, Q) are shown as node, and relationship is expressed as
Side,
Preferably, the construction unit includes:
Second conversion unit, for scenic spot entity, attribute and the attribute value in triple to be converted into digital shape word；
Retrieval unit, for taking out the ID of the sight spot entity in triple, attribute and attribute value；
Normalization unit, for by sight spot entity ID, Property ID and attribute value ID with a low-dimensional real-valued vectors indicate with
The vector for obtaining sight spot entity, attribute and attribute value, is normalized institute's directed quantity；
Training unit, for constructing scoring function, training triple, to update the scape according to the distance properties of triple
The vector of point entity, attribute and attribute value indicates.
As described above, a kind of personalized scape recommended method based on user's positive and negative feedback portrait coding of the invention and being
System, has the advantages that
1, the feature at sight spot is constituted triple by the present invention, and using triple as building unit knowledge mapping.Triple
Using the storage for not only simplifying data, attribute possessed by sight spot is also maintained；The use of knowledge mapping accurately describes scape
The relationship of point and its attribute, so that data have reliability and accuracy.
2, the method that the present invention is learnt using network representation, by the triple vectorization in knowledge mapping, i.e., by triple
Middle entity and relationship map are expressed as vector into low dimensional vector space.This method is by many and diverse network node structure and attribute
Data are expressed as simple vector, not only keep the characteristic of original data, also greatly simplify and calculate, so that recommendation results have
Reliability and reasonability.
3, the present invention has used different weight coefficient and scape when portraying portrait coding of the user to sight spot positive and negative feedback
Point entity vector is calculated, and this method considers user and likes and dislike degree to sight spot, so that portrait description is more sticked on
Nearly individual subscriber actual preference.
4, present invention introduces the positive and negative feedback of user portrait, the positive feedback portrait of user, which is used to calculate, recommends alternative sight spot,
Negative-feedback portrait optimizes positive feedback resulting sight spot of drawing a portrait, to ensure that the accuracy and reasonability of recommending scenery spot.
Detailed description of the invention
In order to which the present invention is further explained, described content, with reference to the accompanying drawing makees a specific embodiment of the invention
Further details of explanation.It should be appreciated that these attached drawings are only used as typical case, and it is not to be taken as to the scope of the present invention
It limits.
Fig. 1 is recommending scenery spot overall structure flow chart；
Fig. 2 is data acquisition and processing (DAP) flow chart；
Fig. 3 is triple vectorization flow chart；
Fig. 4 is that the portrait of the positive and negative feedback of user encodes flow chart；
Fig. 5 is that recommending scenery spot generates flow chart.
Specific embodiment
Illustrate embodiments of the present invention below by way of specific specific example, those skilled in the art can be by this specification
Other advantages and efficacy of the present invention can be easily understood for disclosed content.The present invention can also pass through in addition different specific realities
The mode of applying is embodied or practiced, the various details in this specification can also based on different viewpoints and application, without departing from
Various modifications or alterations are carried out under spirit of the invention.It should be noted that in the absence of conflict, following embodiment and implementation
Feature in example can be combined with each other.
It should be noted that illustrating the basic structure that only the invention is illustrated in a schematic way provided in following embodiment
Think, only shown in schema then with related component in the present invention rather than component count, shape and size when according to actual implementation
Draw, when actual implementation kenel, quantity and the ratio of each component can arbitrarily change for one kind, and its assembly layout kenel
It is likely more complexity.
As shown in Figure 1, the present invention provides a kind of personalized scape recommended method based on user's positive and negative feedback portrait coding, tool
Body the following steps are included:
Step S1, using web crawlers acquisition user to the history evaluation information at sight spot and the Tourism Attribute data at sight spot,
And it is pre-processed；Further according to every user to the height of the figure of merit at sight spot, the positive and negative evaluation sight spot of the user is obtained；Institute
State Tourism Attribute belong to according to include sight spot grade, sight spot scoring, scenic spot type (multinomial), geographical location, ticket price, play when
Length is suitable for season etc. of playing.
Step S2, sight spot and its Tourism Attribute value are indicated with triple.Citing: " Lijiang River scenic spot is national 5A grades of scape
Area " can then use triple<Lijiang River scenic spot, and the rank of the scenic spot is 5A grades national>to indicate, wherein " Lijiang River scenic spot " is sight spot entity,
" the rank of the scenic spot " is the attribute of a relation of triple, " the 5A grades national " attribute for sight spot.By all sight spots and its Tourism Attribute value
The above triple is converted into construct sight spot knowledge mapping；
Step S3, triple is mapped in characteristic vector space using the method that network representation learns, it will be in triple
Sight spot entity, attribute and attribute value be trained by score function, and sight spot entity, attribute and attribute value is made to be converted to vector
The form of expression；
Step S4, according to user to the height of the scoring score value at sight spot, the corresponding weight system of setting sight spot entity vector
Number, then the sight spot step S3 entity vector is calculated with corresponding weight coefficient, respectively acquisition user to sight spot just
Negative-feedback portrait coding；
Step S5, similarity meter is made with the step S3 sight spot entity vector obtained using positive feedback portrait coding in step S4
Calculation obtains sight spot and the reservation that user likes；Negative-feedback portrait coding in recycle step S4 optimizes the sequence, obtains
To final sequence.
As shown in Fig. 2, the acquisition user is to the history evaluation information at sight spot and the Tourism Attribute data at sight spot, and carry out
Pretreatment；Further according to every user to the height of the figure of merit at sight spot, the positive and negative evaluation sight spot of the user is obtained, it is specific to wrap
It includes:
Step S11, using existing web crawlers technology, take journey, with crawling Guilin City on the tour sites such as journey tourism
A variety of Tourism Attributes at interior sight spot and sight spot, Tourism Attribute include: sight spot grade, sight spot scoring, scenic spot type (multinomial),
It manages position, ticket price, duration of playing, be suitable for season etc. of playing.Since the initial data crawled cannot comply fully with we
The requirement of the subsequent calculating of method needs to carry out necessary processing and conversion to it.Such as sight spot geographical location, what is crawled is specific door
The trade mark, but we are address according to administrative division when constructing triple；Such as sight spot admission ticket, what is crawled is 50,90
Or 100 the specific amount of money, but when we save attribute, is divided according to price range, and high, normal, basic third gear is subsequently becomed；Example
If sight spot is scored, what is crawled is the comprehensive score 3.8,4.2 or 4.8 at sight spot, is rounded up when saving comprehensive score to score value,
Subsequently become 1 point, 2 points, 3 points, 4 points and 5 points five grades.Meanwhile every user is crawled on website to sight spots all in Guilin City
Practical score information.
Step S12, it since the step S11 initial data crawled cannot be calculated directly, needs for the setting of these data
Unique ID value.Data in step S11 are divided into four class data by this method, and with four tables store respectively these four types of data and
Corresponding ID value.Four tables are respectively user-id table, sight spot ID table, Property ID table and attribute value ID table, and wherein user-id table is deposited
Storage is user's name and corresponding ID value, and ID table storage in sight spot is sight name and corresponding ID value, the storage of Property ID table
Be Property Name and corresponding ID value, the storage of attribute value ID table be sight spot all properties value title and corresponding ID
Value；In this four tables, the ID value of first data is set as 1, and the ID value of follow-up data adds 1 one by one, such as in the ID table of sight spot,
The ID value at " Lijiang River scenic spot " is 1, and the ID value at " Elephant Trunk Hill park scenic spot " is 2, and the ID value at " Seven Star Park scenic spot " is 3；Such as
In attribute value ID table, the ID value at " 5A grades of scenic spots " is 1, and the ID value at " 4A grades of scenic spots " is 2, and the ID value at " 3A grades of scenic spots " is 3；
Step S13, since the collected user of institute scoring practical to sight spot is not necessarily all integer, score value is all at 0 point and 5
/, in order to which accurate description user is to the characteristic liked and disliked at sight spot, these score values handle as follows: greater than 0
Dividing and being less than or equal to 1 point is 1 point, and being greater than 1 point and being less than or equal to 2 points is 2 points, and user is divided into the scoring at sight spot by rule according to this
Five grades: 5 points, 4 points, 3 points, 2 points, 1 point；Wherein the sight spot more than or equal to 3 points is the sight spot that user likes, less than 3 points
Sight spot is that user dislikes sight spot；And using list score of the user to sight spot is stored as user respectively and the positive evaluation at sight spot is arranged
Table and negative evaluation list.
In this present embodiment, described to convert triple for all sight spots and its Tourism Attribute value to construct sight spot knowledge graph
Spectrum, specifically:
The form that the data triple such as a variety of attributes at sight spot and sight spot (P, V, Q) will be collected indicates that wherein P is scape
Point entity, V are attributes, and Q is the attribute value at scenic spot, such as: Lijiang River scenic spot is the national scenic spot 5A grades of, wherein " Lijiang River scenic spot " indicates
For sight spot entity P, " 5A grades of scenic spots " indicates attribute value Q, and the relationship between " Lijiang River scenic spot " and " 5A grades of scenic spots " indicates attribute V,
And the triple constructed is stored using Neo4j database.Using such representation method, scape can effectively be described
The feature of point meets user to the personalized demand of sight spot attributive character.The sight spot entity and attribute value table of triple are shown as
Node, relationship are expressed as side, and according to the property of triple, two nodes are connected using side, constitute a net about sight spot
Network, the i.e. knowledge mapping at sight spot.
Since the representation of triple not can be carried out effective similarity calculation, the portrait structure of user can not be also carried out
It builds, the present invention needs for entity in triple and attribute of a relation to be converted into the representation of vector.This example uses network representation
Triple is converted the form of vector by the TransE model in study.As shown in figure 3, the method learnt using network representation
Triple in the knowledge mapping of sight spot is mapped in characteristic vector space, by sight spot entity, attribute and the attribute in triple
Value is trained by score function, so that sight spot entity, attribute and attribute value is converted to vector representation, is specifically included:
Step S31, from the sight spot entity taken out in knowledge mapping in triple, attribute and attribute value, then according to step
The sight spot ID table of S12, Property ID table and attribute value ID table, by the sight spot entity of triple, attribute and attribute value are converted into number
Form, and using in the text of entitled Triple, to treated, triple is stored, each triple in the text
A line is accounted for, such as " Elephant Trunk Hill park scenic spot is located at Guilin City Qixing District " is converted into being expressed as (P1, V10, Q9) after number, in text
The first row is accounted in this；" Elephant Trunk Hill park scenic spot belongs to 5A grades of scenic spots " is converted into being expressed as (P1, V1, Q1) after number, in text
In account for the second row.
Step S32, sight spot entity ID, Property ID and the attribute value ID of triple are taken out in knowledge mapping.
Step S33, each entity and attribute value ID will be indicated with a low-dimensional real-valued vectors in step S32, model
Initially by each dimension of random value initialization vector of the vector normal distribution, sight spot entity vector, attribute vector can get
With attribute value vector, 100 dimensions are may be selected in the dimension of this example；Then all vectors are normalized, by being normalized to
The mould length of dominant vector.
Step S34, sight spot entity vector P, attribute vector V and the attribute value vector Q after normalization are trained, make this
The feature vector that three vectors meet this distance properties of P+V=Q namely entity P in vector space adds the vector of attribute V
Equal to the feature vector of attribute value Q.Wherein, it trains triple and is mapped to the scoring function of vector space are as follows:
fVIndicate scoring function, L1/2Indicate that L can be used in scoring function1Normal form or L2Normal form carries out operation, entire public
Formula is meant that sight spot entity vector P is added resulting vector approach attribute value vector Q with attribute vector V.It is used in training
The vector that the method for stochastic gradient descent updates sight spot entity, attribute and attribute value indicates, so that the vector of three objects calculates
Meet scoring function fVRequirement.
During actual model training, in order to distinguish the triple of right and wrong, used objective function are as follows:
Wherein, S is the set of correct triple, S-It is the set of wrong triple, in the x and y that max (x, y) is returned
Biggish value, γ are the spacing distance between correct triple score and the score of wrong triple.TransE model training
Process is namely by adjusting the feature vector of entity, attribute and attribute value in each triple, so that objective function L is obtained most
Greatly, namely by optimizing feature vector make positive example triple score as close possible to 0, and negative example triple score is as inclined as possible
From 0.By the calculating of majorized function, so that the entity vector P in positive example triple is added resulting vector with attribute vector V and gets over
To be added resulting vector increasingly with attribute vector V closer to attribute value vector Q, the sight spot entity vector P of negative example triple
Far from attribute value vector Q.
It positive example triple in TransE model and is made of the objective fact in the knowledge mapping of sight spot, negative example triple is then
It is obtained by algorithm by the attribute value in replacement positive example triple, that is, constitutes the triple opposite with objective fact.Such as positive example
Triple<Lijiang River scenic spot, the rank of the scenic spot are 5A grades national>can to construct negative example triple<Lijiang River scenic spot, the rank of the scenic spot, national 2A
Grade>,<Lijiang River scenic spot, the rank of the scenic spot, it is 3A grades national>etc..Model is by constructing several negative example triples and cooperating positive example ternary
Group completes the training of each sight spot feature vector in the knowledge mapping of sight spot.
After the completion of being mapped to vector space and training due to sight spot entity, attribute and the attribute value of triple, sight spot entity
Vector, the structure between attribute vector and attribute value vector is constant, and entity vector corresponding attribute value vector in sight spot still retains this
The attributive character at sight spot, therefore scenic spot vector P still has this attributive character of attribute value vector V, attribute value vector can be very well
The corresponding sight spot attributive character of expression；If two different sight spot entities have same attribute value, pass through TransE model
Two entity vectors after training have similitude in vector space；Likewise, two different sight spots are with a variety of similar
Attribute value, by two sight spot entity vectors after TransE model training there is also the relationship of similitude in vector space,
Citing, such as type of tour, admission fee, the duration of playing of Seven Star Park and Elephant Trunk Hill park etc. have similitude, therefore the two
Sight spot also has similitude in knowledge mapping, and corresponding sight spot entity vector is also more similar in vector space.As long as
All sight spot entity vectors liked and disliked of user are calculated, just can depict what user liked and disliked to the characteristic at sight spot
Portrait.
Each user can go sight-seeing different sight spots, also different to the sight spot scoring after visit, and this makes it possible to obtain users
It is also different to the portrait at sight spot.In the present system, the portrait coding that the sight spot that user likes is constituted is known as positive feedback portrait coding,
The portrait coding that the sight spot that user dislikes is constituted is known as negative-feedback portrait coding.The positive and negative feedback portrait coding flow chart of user is such as
Shown in Fig. 4, specific step and it is described below:
Step S41, the sight spot entity vector of positive and negative evaluation is obtained respectively.
Since sight spot ID is unique, according to certain user to sight spot positive and negative evaluation list, the user obtained respectively likes
With disagreeable sight spot ID, the corresponding sight spot entity vector of two evaluation lists is then obtained respectively from knowledge mapping；
Step S42, the weight coefficient of the scape entity vector of positive and negative evaluation is respectively set.
The scoring that user likes and dislike sight spot is obtained respectively to the positive and negative evaluation list at sight spot by user, according to user
The score value at the sight spot liked, setting sight spot entity vector draw a portrait in positive feedback and encode corresponding weight coefficient in calculating, according to
The score value at the sight spot that user dislikes, setting sight spot entity vector draw a portrait in negative-feedback and encode corresponding weight coefficient in calculating；
Wherein, the weight coefficient for the sight spot vector just evaluated is provided that the weight coefficient that 5 points of weight coefficient is Isosorbide-5-Nitrae point is 0.6,3
The weight coefficient divided is 0.4, and the sight spot vector weight coefficient of negative evaluation is provided that 1 point of weight coefficient for 0.8,2 points
Weight coefficient is 0.2.This method for assigning different weight coefficients to positive and negative evaluation sight spot vector, it is identical relative to using
The calculation method of weight can more accurate description user preference, and then the better user-customized recommended result of generation.
Step S43, positive and negative portrait is obtained according to the sight spot entity vector of positive and negative evaluation and respective weights system-computed to encode.
Positive and negative feedback portrait coding is to carry out calculating acquisition by the corresponding weight of sight spot entity vector sum.Positive feedback weight
Use p1、p2、p31,0.8 and 0.5 is respectively indicated, negative-feedback weight n1、n20.8 and 0.2 are respectively indicated, sight spot entity vector uses
M expression,It is expressed as user and likes j-th of sight spot entity vector,Indicate that user dislikes j-th of sight spot entity vector, user
The number at the sight spot liked indicates that the number at the sight spot that user dislikes is indicated using M using N.By the sight spot vector and phase of user
The product accumulation of corresponding weight coefficient obtains the user and encodes to the portrait of sight spot positive and negative feedback, and wherein positive feedback portrait is compiled
Shown in the calculating such as formula (3) of code:
Shown in the calculating such as formula (4) of negative-feedback portrait coding:
In formula (3),What is indicated is the portrait coding of positive feedback of k-th of user to sight spot；In formula (4),What is indicated is the portrait coding of negative-feedback of k-th of user to sight spot.
According to expression formula two kinds of users obtained draw a portrait coding play the role of it is different, the positive feedback portrait coding of user
The attributive character that the sight spot that user likes has is described, similar sight spot is encoded with the portrait to get use is arrived by calculating
The sight spot that family is liked；Negative Feedback Coding and description is attributive character that sight spot that user dislikes has, is drawn by negative-feedback
The recommendation sight spot that positive feedback portrait coding generates is advanced optimized as coding, deletes the similar sight spot that wherein user dislikes to mention
The accuracy of high result.Similarity calculation, which is done, with sight spot entity vector using positive feedback portrait coding obtains the scape that user likes
Point；It recycles negative-feedback portrait coding to make similarity calculation with sight spot entity vector, is finally sorted.As shown in figure 5, specific
The step of include:
Step S51, it is drawn a portrait using positive feedback and encodes the positive feedback that calculating obtains the alternative recommendation each user in sight spot of user
Portrait coding makees similarity calculation with all sight spot entity vectors in knowledge mapping, obtains each user and likes sight spot, still
There may be the sight spot that user had gone sight-seeing among these sight spots, and recommending scenery spot algorithm mainly is recommended not going to user
The sight spot crossed, it is therefore desirable to the sight spot gone sight-seeing be deleted, the sight spot reservation of K before ranking is then selected；
Step S52, similarity is carried out using negative-feedback portrait coding and the sight spot entity vector of K before ranking in step S51
It calculates, obtains the sights that each user does not like；
Step S53, K/2 scape before the ranking that each user least likes is deleted before ranking in the sight spot of K in step s 51
Point, remaining sight spot are the sight spot to be recommended of this example.
Pearson came, theorem in Euclid space, cosine similarity etc. can be used in the similarity calculating method of step S51 and step S52
The method of similarity.It is calculated in the present embodiment using cosine similarity method.Known by formula (3)What is indicated is kth
A user encodes the portrait of the positive feedback at sight spot, MuIt is u-th of sight spot entity vector, the calculating of their cosine similarity is such as
Shown in formula (5):
It is expressed as the similarity of the portrait coding of the positive feedback of sight spot u and user k.By the portrait of positive feedback coding with
All sight spot entity vectors make first time similarity calculation, and the interested sight spot of each user can be obtained, then select K before ranking
Sight spot, it is that the 10 specific recommendation lists of explanation generate process that this example, which selects K,.Calculating all user's positive feedback portrait codings and scape
After point similarity, preceding 10 sight spot is ranked up from high to low according to similarity size, the result of sequence is to recommend alternative scape
Point；It recycles 10 sight spot entity vectors of negative-feedback portrait coding and this to carry out second of similarity calculation, is known by formula (4)
What is indicated is the portrait coding of negative-feedback of k-th of user to sight spot, the calculating of second of cosine similarity such as formula (6) institute
Show:
Wv -The negative-feedback similarity for indicating the portrait coding of v-th of sight spot and the negative-feedback of user k, then according to similarity
Size this 10 sight spots are ranked up, as a result in ranking before 55 sight spots least liking as user of sight spot, will
This 5 sight spots are deleted from 10 sight spots that positive feedback portrait has sorted, and being left 5 sight spots is the sight spot that this example is recommended.
The present invention is encoded by negative-feedback portrait of the user to sight spot feature, and on the basis of favorite interest list
Increase the sight spot similarity calculation that a user dislikes and optimize and obtain consequently recommended list, effectively user is liked and disliked
Sight spot distinguish, on the basis of meeting user individual, provide more accurate recommending scenery spot for user.
Embodiment two
A kind of personalized recommending scenery spot system based on user's positive and negative feedback portrait coding, the system include:
Data acquisition and procession unit, for acquiring user to the history evaluation information at sight spot and the Tourism Attribute number at sight spot
According to, and pre-processed；Further according to every user to the height of the figure of merit at sight spot, the positive and negative evaluation scape of the user is obtained
Point；Converting unit, for converting triple for all sight spots and its Tourism Attribute value to construct sight spot knowledge mapping；
Triple in the knowledge mapping of sight spot is mapped to feature by construction unit, the method for being learnt by network representation
In vector space, sight spot entity, attribute and the attribute value in triple are trained by score function, make sight spot entity,
Attribute and attribute value are converted to vector representation；
Sight spot entity is arranged for the height according to user to the figure of merit at sight spot in positive and negative feedback portrait coding unit
Then sight spot entity vector is calculated with corresponding weight coefficient, is used respectively by the corresponding weight coefficient of vector
It draws a portrait and encodes to the positive and negative feedback at sight spot in family；
Recommendation unit obtains user for sight spot entity vector doing similarity calculation using positive feedback portrait coding and likes
Sight spot, and sight spot is ranked up from high to low according to similarity；Negative-feedback portrait coding is recycled to carry out the sequence excellent
Change, is finally sorted.
In this present embodiment, the data acquisition and procession unit includes:
Data acquisition unit, for acquiring user to the history evaluation information at sight spot and the Tourism Attribute data at sight spot；
Configuration unit, for ID to be arranged for the attribute value data at user, sight spot and sight spot；
Division unit likes sight spot and disagreeable scape for being divided into sight spot according to history evaluation information of the user to sight spot
Point.
In this present embodiment, the converting unit includes:
First conversion unit, the shape of the Tourism Attribute data triple (P, V, Q) for sight spot and sight spot will to be collected
Formula indicates that wherein P is sight spot entity, and V is attribute, and Q is sight spot attribute value；
Knowledge mapping acquiring unit constitutes a network about sight spot, i.e., for two nodes of triple to be connected
The knowledge mapping at sight spot, wherein the sight spot entity and sight spot attribute value table of triple (P, V, Q) are shown as node, and relationship is expressed as
Side,
In this present embodiment, the construction unit includes:
Second conversion unit, for scenic spot entity, attribute and the attribute value in triple to be converted into digital shape word；
Retrieval unit, for taking out the ID of the sight spot entity in triple, attribute and attribute value；
Normalization unit, for by sight spot entity ID, Property ID and attribute value ID with a low-dimensional real-valued vectors indicate with
The vector for obtaining sight spot entity, attribute and attribute value, is normalized institute's directed quantity；
Training unit, for constructing scoring function, training triple, to update the scape according to the distance properties of triple
The vector of point entity, attribute and attribute value indicates.
In this present embodiment, the positive and negative feedback portrait coding unit includes:
Sight spot entity vector acquiring unit, for obtaining the sight spot entity vector of positive and negative evaluation；
Weight coefficient setting unit, the weight coefficient of the scape entity vector for positive and negative evaluation to be arranged；
Positive and negative portrait encodes computing unit, for the sight spot entity vector and respective weights system-computed according to positive and negative evaluation
Obtain positive and negative portrait coding.
In this present embodiment, the recommendation unit includes:
Alternative sight spot acquiring unit obtains the alternative recommendation sight spot of user for drawing a portrait to encode to calculate according to positive feedback；
Sight spot acquiring unit is not liked, for the sight spot for encoding and calculating and obtaining user and do not like of drawing a portrait according to negative-feedback；
Alternative recommending scenery spot unit obtains for the recommendation sight spot alternative according to negative-feedback portrait code optimization and recommends scape
Point.
In the present embodiment, the specific method of the function of all realizations of each unit can be real by method above-mentioned
Existing, the present embodiment is just not repeated.
The present invention is encoded by negative-feedback portrait of the user to sight spot feature, and on the basis of favorite interest list
Increase the sight spot similarity calculation that a user dislikes and optimize and obtain consequently recommended list, effectively user is liked and disliked
Sight spot distinguish, on the basis of meeting user individual, provide more accurate recommending scenery spot for user.
The above-described embodiments merely illustrate the principles and effects of the present invention, and is not intended to limit the present invention.It is any ripe
The personage for knowing this technology all without departing from the spirit and scope of the present invention, carries out modifications and changes to above-described embodiment.Cause
This, institute is complete without departing from the spirit and technical ideas disclosed in the present invention by those of ordinary skill in the art such as
At all equivalent modifications or change, should be covered by the claims of the present invention.
Claims (10)
1. the personalized recommending scenery spot method based on user's positive and negative feedback portrait coding, which is characterized in that this method includes following
Step:
User is acquired to the history evaluation information at sight spot and the Tourism Attribute data at sight spot, and is pre-processed；Further according to every
User obtains the positive and negative evaluation sight spot of the user to the height of the figure of merit at sight spot；
Triple is converted by all sight spots and its Tourism Attribute value to construct sight spot knowledge mapping；
The triple in the knowledge mapping of sight spot is mapped in characteristic vector space using the method that network representation learns, by ternary
Sight spot entity, attribute and attribute value in group are trained by score function, are converted to sight spot entity, attribute and attribute value
Vector representation；
According to user to the height of the figure of merit at sight spot, the corresponding weight coefficient of sight spot entity vector is set, then by scape
Point entity vector is calculated with corresponding weight coefficient, obtains user respectively to the positive and negative feedback portrait coding at sight spot；
Similarity calculation is done with sight spot entity vector using positive feedback portrait coding and obtains the sight spot that user likes, and according to similar
Degree is from high to low ranked up sight spot；It recycles negative-feedback portrait coding to optimize the sequence, is finally sorted.
2. the personalized recommending scenery spot method according to claim 1 based on user's positive and negative feedback portrait coding, feature
It is, the acquisition user pre-processes the history evaluation information at sight spot and the Tourism Attribute data at sight spot；Further according to
Every user obtains the positive and negative evaluation sight spot of the user, specifically includes to the height of the figure of merit at sight spot:
User is acquired to the history evaluation information at sight spot and the Tourism Attribute data at sight spot；
For the attribute value data at user, sight spot and sight spot, ID is set；
Sight spot is divided into according to history evaluation information of the user to sight spot and likes sight spot and disagreeable sight spot.
3. the personalized recommending scenery spot method according to claim 1 based on user's positive and negative feedback portrait coding, feature
It is, it is described to convert triple for all sight spots and its Tourism Attribute value to construct sight spot knowledge mapping, it specifically includes:
The form that the Tourism Attribute data triple (P, V, Q) at sight spot and sight spot will be collected indicates that wherein P is sight spot reality
Body, V are attributes, and Q is sight spot attribute value；The sight spot entity and sight spot attribute value table of triple (P, V, Q) are shown as node, relation table
It is shown as side, two nodes are connected using side, constitutes a network about sight spot, the i.e. knowledge mapping at sight spot.
4. the personalized recommending scenery spot method according to claim 1 based on user's positive and negative feedback portrait coding, feature
It is, the triple in the knowledge mapping of sight spot is mapped in characteristic vector space by the method learnt using network representation,
Sight spot entity, attribute and attribute value in triple is trained by score function, makes sight spot entity, attribute and attribute value
It is converted to vector representation, is specifically included:
Scenic spot entity, attribute and attribute value in triple is converted into digital shape word；
The ID of sight spot entity, attribute and attribute value in taking-up triple；
Sight spot entity ID, Property ID and attribute value ID are indicated with a low-dimensional real-valued vectors with obtain sight spot entity, attribute and
Institute's directed quantity is normalized in the vector of attribute value；
Scoring function, training triple, to update the sight spot entity, attribute and attribute are constructed according to the distance properties of triple
The vector of value indicates.
5. the personalized recommending scenery spot method according to claim 1 based on user's positive and negative feedback portrait coding, feature
Be, it is described according to user to the height of the figure of merit at sight spot, the corresponding weight coefficient of sight spot entity vector is set, then
Sight spot entity vector is calculated with corresponding weight coefficient, user is obtained respectively and the positive and negative feedback portrait at sight spot is compiled
Code, specifically includes:
The sight spot entity vector of positive and negative evaluation is obtained respectively；
The weight coefficient of the sight spot entity vector of positive and negative evaluation is respectively set；
It is calculated according to the sight spot entity vector of positive and negative evaluation and respective weights coefficient and obtains positive and negative feedback portrait coding.
6. the personalized recommending scenery spot method according to claim 1 based on user's positive and negative feedback portrait coding, feature
It is, it is described to do similarity calculation using positive feedback portrait coding and sight spot entity vector and obtain the sight spot that user likes, and root
Sight spot is ranked up from high to low according to similarity；It recycles negative-feedback portrait coding to optimize the sequence, obtains final
Sequence, specifically includes:
It draws a portrait to encode to calculate using positive feedback and obtains the alternative recommendation sight spot of user；
It is drawn a portrait using negative-feedback and encodes the sight spot that calculating acquisition user does not like；
Using the alternative recommendation sight spot of negative-feedback portrait code optimization, obtains and recommend sight spot.
7. the personalized recommending scenery spot system based on user's positive and negative feedback portrait coding, which is characterized in that the system includes:
Data acquisition and procession unit, for acquiring user to the history evaluation information at sight spot and the Tourism Attribute data at sight spot,
And it is pre-processed；Further according to every user to the height of the figure of merit at sight spot, the positive and negative evaluation sight spot of the user is obtained；
Converting unit, for converting triple for all sight spots and its Tourism Attribute value to construct sight spot knowledge mapping；
Triple in the knowledge mapping of sight spot is mapped to feature vector by construction unit, the method for being learnt by network representation
In space, sight spot entity, attribute and the attribute value in triple are trained by score function, make sight spot entity, attribute
Vector representation is converted to attribute value；
Sight spot entity vector is arranged for the height according to user to the figure of merit at sight spot in positive and negative feedback portrait coding unit
Then corresponding weight coefficient calculates sight spot entity vector with corresponding weight coefficient, obtain user couple respectively
The positive and negative feedback at sight spot, which is drawn a portrait, to be encoded；
Recommendation unit obtains the scape that user likes for doing similarity calculation with sight spot entity vector using positive feedback portrait coding
Point, and sight spot is ranked up from high to low according to similarity；It recycles negative-feedback portrait coding to optimize the sequence, obtains
To final sequence.
8. the personalized recommending scenery spot system according to claim 7 based on user's positive and negative feedback portrait coding, feature
It is, the data acquisition and procession unit includes:
Data acquisition unit, for acquiring user to the history evaluation information at sight spot and the Tourism Attribute data at sight spot；
Configuration unit, for ID to be arranged for the attribute value data at user, sight spot and sight spot；
Division unit likes sight spot and disagreeable sight spot for being divided into sight spot according to history evaluation information of the user to sight spot.
9. the personalized recommending scenery spot system according to claim 7 based on user's positive and negative feedback portrait coding, feature
It is, the converting unit includes:
First conversion unit, the form table of the Tourism Attribute data triple (P, V, Q) for sight spot and sight spot will to be collected
Show, wherein P is sight spot entity, and V is attribute, and Q is sight spot attribute value；
Knowledge mapping acquiring unit constitutes a network about sight spot, i.e. sight spot for two nodes of triple to be connected
Knowledge mapping, wherein the sight spot entity and sight spot attribute value table of triple (P, V, Q) are shown as node, and relationship is expressed as side.
10. the personalized recommending scenery spot system according to claim 7 based on user's positive and negative feedback portrait coding, feature
It is, the construction unit includes:
Second conversion unit, for scenic spot entity, attribute and the attribute value in triple to be converted into digital shape word；
Retrieval unit, for taking out the ID of the sight spot entity in triple, attribute and attribute value；
Normalization unit, for being indicated sight spot entity ID, Property ID and attribute value ID with a low-dimensional real-valued vectors to obtain
Institute's directed quantity is normalized in the vector of sight spot entity, attribute and attribute value；
Training unit, for constructing scoring function according to the distance properties of triple, training triple is real to update the sight spot
The vector of body, attribute and attribute value indicates.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811131075.8A CN109189944A (en) | 2018-09-27 | 2018-09-27 | Personalized recommending scenery spot method and system based on user's positive and negative feedback portrait coding |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811131075.8A CN109189944A (en) | 2018-09-27 | 2018-09-27 | Personalized recommending scenery spot method and system based on user's positive and negative feedback portrait coding |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109189944A true CN109189944A (en) | 2019-01-11 |
Family
ID=64907507
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811131075.8A Pending CN109189944A (en) | 2018-09-27 | 2018-09-27 | Personalized recommending scenery spot method and system based on user's positive and negative feedback portrait coding |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109189944A (en) |
Cited By (24)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110135888A (en) * | 2019-04-12 | 2019-08-16 | 平安科技（深圳）有限公司 | Product information method for pushing, device, computer equipment and storage medium |
| CN110210988A (en) * | 2019-05-31 | 2019-09-06 | 北京理工大学 | A kind of symbol social networks embedding grammar based on depth Hash |
| CN110287336A (en) * | 2019-06-19 | 2019-09-27 | 桂林电子科技大学 | A kind of tourist's portrait construction method recommended towards tourist attractions |
| CN110288436A (en) * | 2019-06-19 | 2019-09-27 | 桂林电子科技大学 | A kind of personalized recommending scenery spot method based on the modeling of tourist's preference |
| CN110287335A (en) * | 2019-06-17 | 2019-09-27 | 桂林电子科技大学 | The personalized recommending scenery spot method and device of knowledge based map and user's shot and long term preference |
| CN110457596A (en) * | 2019-08-01 | 2019-11-15 | 北京明略软件系统有限公司 | A kind of resource recommendation processing method and processing device |
| CN110472155A (en) * | 2019-07-03 | 2019-11-19 | 五邑大学 | Collaborative recommendation method, device, equipment and the storage medium of knowledge based map |
| CN110532464A (en) * | 2019-08-13 | 2019-12-03 | 桂林电子科技大学 | A kind of tourism recommended method based on more tourism context modelings |
| CN110555112A (en) * | 2019-08-22 | 2019-12-10 | 桂林电子科技大学 | interest point recommendation method based on user positive and negative preference learning |
| CN111161009A (en) * | 2019-11-19 | 2020-05-15 | 中国电子产品可靠性与环境试验研究所（（工业和信息化部电子第五研究所）（中国赛宝实验室）） | Information pushing method and device, computer equipment and storage medium |
| CN111198988A (en) * | 2019-12-25 | 2020-05-26 | 中国平安财产保险股份有限公司 | Service recommendation method, training method, device, computer equipment and storage medium |
| CN111507796A (en) * | 2020-04-07 | 2020-08-07 | 桂林电子科技大学 | Online shopping mall commodity recommendation method based on knowledge graph |
| CN111949885A (en) * | 2020-08-27 | 2020-11-17 | 桂林电子科技大学 | Personalized recommendation method for scenic spots |
| CN112232408A (en) * | 2020-10-15 | 2021-01-15 | 平安科技（深圳）有限公司 | Target recommendation method and device, electronic equipment and computer-readable storage medium |
| CN112287043A (en) * | 2020-12-29 | 2021-01-29 | 成都数联铭品科技有限公司 | Automatic graph code generation method and system based on domain knowledge and electronic equipment |
| WO2021031480A1 (en) * | 2019-08-21 | 2021-02-25 | 广州视源电子科技股份有限公司 | Text generation method and device |
| CN112992021A (en) * | 2021-03-08 | 2021-06-18 | 合肥工业大学 | Intelligent scenic spot navigation system and use method |
| CN113127750A (en) * | 2021-05-19 | 2021-07-16 | 同盾科技有限公司 | Information list generation method and device, storage medium and electronic equipment |
| CN113254843A (en) * | 2021-06-29 | 2021-08-13 | 腾讯科技（深圳）有限公司 | Information pushing method and device and storage medium |
| CN113342931A (en) * | 2021-05-27 | 2021-09-03 | 东风柳州汽车有限公司 | Big data based user demand analysis method, device, equipment and storage medium |
| CN113902354A (en) * | 2021-12-03 | 2022-01-07 | 环球数科集团有限公司 | Travel evaluation data processing method and device and computer equipment |
| US20220019341A1 (en) * | 2020-07-14 | 2022-01-20 | Beijing Baidu Netcom Science And Technology Co., Ltd. | Map information display method and apparatus, electronic device, and computer storage medium |
| CN114117233A (en) * | 2021-12-03 | 2022-03-01 | 上海交通大学 | Conversation news recommendation method and system based on user implicit feedback |
| JP2023091701A (en) * | 2021-12-20 | 2023-06-30 | ネイバー コーポレーション | Method for place personalization recommendation based on user preference, computer system, and computer program |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150317398A1 (en) * | 2010-12-30 | 2015-11-05 | Google Inc. | Presenting non-suggested content items to a user of a social network account |
| CN107273490A (en) * | 2017-06-14 | 2017-10-20 | 北京工业大学 | A kind of combination mistake topic recommendation method of knowledge based collection of illustrative plates |
| CN107507049A (en) * | 2017-06-30 | 2017-12-22 | 昆明理工大学 | Method is recommended in a kind of online service towards inconsistent user's interpretational criteria |
| CN107729444A (en) * | 2017-09-30 | 2018-02-23 | 桂林电子科技大学 | Recommend method in a kind of personalized tourist attractions of knowledge based collection of illustrative plates |
| CN107862022A (en) * | 2017-10-31 | 2018-03-30 | 中国科学院自动化研究所 | Cultural resource commending system |
-
2018
- 2018-09-27 CN CN201811131075.8A patent/CN109189944A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150317398A1 (en) * | 2010-12-30 | 2015-11-05 | Google Inc. | Presenting non-suggested content items to a user of a social network account |
| CN107273490A (en) * | 2017-06-14 | 2017-10-20 | 北京工业大学 | A kind of combination mistake topic recommendation method of knowledge based collection of illustrative plates |
| CN107507049A (en) * | 2017-06-30 | 2017-12-22 | 昆明理工大学 | Method is recommended in a kind of online service towards inconsistent user's interpretational criteria |
| CN107729444A (en) * | 2017-09-30 | 2018-02-23 | 桂林电子科技大学 | Recommend method in a kind of personalized tourist attractions of knowledge based collection of illustrative plates |
| CN107862022A (en) * | 2017-10-31 | 2018-03-30 | 中国科学院自动化研究所 | Cultural resource commending system |
Non-Patent Citations (1)
| Title |
|---|
| 唐慧琳: "融合结构和语义信息的知识图谱补全算法研究", 《中国优秀硕士学位论文全文数据库信息科技辑（月刊 ）》 * |
Cited By (38)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110135888A (en) * | 2019-04-12 | 2019-08-16 | 平安科技（深圳）有限公司 | Product information method for pushing, device, computer equipment and storage medium |
| CN110135888B (en) * | 2019-04-12 | 2023-08-08 | 平安科技（深圳）有限公司 | Product information pushing method, device, computer equipment and storage medium |
| CN110210988A (en) * | 2019-05-31 | 2019-09-06 | 北京理工大学 | A kind of symbol social networks embedding grammar based on depth Hash |
| CN110210988B (en) * | 2019-05-31 | 2021-04-27 | 北京理工大学 | Symbolic social network embedding method based on deep hash |
| CN110287335A (en) * | 2019-06-17 | 2019-09-27 | 桂林电子科技大学 | The personalized recommending scenery spot method and device of knowledge based map and user's shot and long term preference |
| CN110288436A (en) * | 2019-06-19 | 2019-09-27 | 桂林电子科技大学 | A kind of personalized recommending scenery spot method based on the modeling of tourist's preference |
| CN110287336B (en) * | 2019-06-19 | 2021-08-27 | 桂林电子科技大学 | Tourist map construction method for tourist attraction recommendation |
| CN110287336A (en) * | 2019-06-19 | 2019-09-27 | 桂林电子科技大学 | A kind of tourist's portrait construction method recommended towards tourist attractions |
| CN110472155A (en) * | 2019-07-03 | 2019-11-19 | 五邑大学 | Collaborative recommendation method, device, equipment and the storage medium of knowledge based map |
| CN110457596A (en) * | 2019-08-01 | 2019-11-15 | 北京明略软件系统有限公司 | A kind of resource recommendation processing method and processing device |
| CN110532464A (en) * | 2019-08-13 | 2019-12-03 | 桂林电子科技大学 | A kind of tourism recommended method based on more tourism context modelings |
| CN110532464B (en) * | 2019-08-13 | 2022-04-12 | 桂林电子科技大学 | Tourism recommendation method based on multi-tourism context modeling |
| WO2021031480A1 (en) * | 2019-08-21 | 2021-02-25 | 广州视源电子科技股份有限公司 | Text generation method and device |
| CN110555112A (en) * | 2019-08-22 | 2019-12-10 | 桂林电子科技大学 | interest point recommendation method based on user positive and negative preference learning |
| CN110555112B (en) * | 2019-08-22 | 2022-07-15 | 桂林电子科技大学 | Interest point recommendation method based on user positive and negative preference learning |
| CN111161009A (en) * | 2019-11-19 | 2020-05-15 | 中国电子产品可靠性与环境试验研究所（（工业和信息化部电子第五研究所）（中国赛宝实验室）） | Information pushing method and device, computer equipment and storage medium |
| CN111161009B (en) * | 2019-11-19 | 2023-11-07 | 中国电子产品可靠性与环境试验研究所（（工业和信息化部电子第五研究所）（中国赛宝实验室）） | Information pushing method, device, computer equipment and storage medium |
| CN111198988A (en) * | 2019-12-25 | 2020-05-26 | 中国平安财产保险股份有限公司 | Service recommendation method, training method, device, computer equipment and storage medium |
| CN111507796A (en) * | 2020-04-07 | 2020-08-07 | 桂林电子科技大学 | Online shopping mall commodity recommendation method based on knowledge graph |
| US20220019341A1 (en) * | 2020-07-14 | 2022-01-20 | Beijing Baidu Netcom Science And Technology Co., Ltd. | Map information display method and apparatus, electronic device, and computer storage medium |
| US11630560B2 (en) * | 2020-07-14 | 2023-04-18 | Beijing Baidu Netcom Science And Technology Co., Ltd. | Map information display method and apparatus, electronic device, and computer storage medium |
| KR102520046B1 (en) * | 2020-07-14 | 2023-04-10 | 베이징 바이두 넷컴 사이언스 앤 테크놀로지 코., 엘티디. | Map information display method, device, electronic equipment and storage medium |
| KR20220008767A (en) * | 2020-07-14 | 2022-01-21 | 베이징 바이두 넷컴 사이언스 앤 테크놀로지 코., 엘티디. | Map information display method, device, electronic equipment and storage medium |
| CN111949885A (en) * | 2020-08-27 | 2020-11-17 | 桂林电子科技大学 | Personalized recommendation method for scenic spots |
| CN112232408A (en) * | 2020-10-15 | 2021-01-15 | 平安科技（深圳）有限公司 | Target recommendation method and device, electronic equipment and computer-readable storage medium |
| CN112232408B (en) * | 2020-10-15 | 2024-06-28 | 平安科技（深圳）有限公司 | Target recommendation method, device, electronic equipment and computer readable storage medium |
| CN112287043B (en) * | 2020-12-29 | 2021-06-18 | 成都数联铭品科技有限公司 | Automatic graph code generation method and system based on domain knowledge and electronic equipment |
| CN112287043A (en) * | 2020-12-29 | 2021-01-29 | 成都数联铭品科技有限公司 | Automatic graph code generation method and system based on domain knowledge and electronic equipment |
| CN112992021A (en) * | 2021-03-08 | 2021-06-18 | 合肥工业大学 | Intelligent scenic spot navigation system and use method |
| CN113127750A (en) * | 2021-05-19 | 2021-07-16 | 同盾科技有限公司 | Information list generation method and device, storage medium and electronic equipment |
| CN113127750B (en) * | 2021-05-19 | 2024-04-02 | 同盾科技有限公司 | Information list generation method and device, storage medium and electronic equipment |
| CN113342931A (en) * | 2021-05-27 | 2021-09-03 | 东风柳州汽车有限公司 | Big data based user demand analysis method, device, equipment and storage medium |
| CN113342931B (en) * | 2021-05-27 | 2022-11-01 | 东风柳州汽车有限公司 | Big data based user demand analysis method, device, equipment and storage medium |
| CN113254843A (en) * | 2021-06-29 | 2021-08-13 | 腾讯科技（深圳）有限公司 | Information pushing method and device and storage medium |
| CN113902354A (en) * | 2021-12-03 | 2022-01-07 | 环球数科集团有限公司 | Travel evaluation data processing method and device and computer equipment |
| CN114117233A (en) * | 2021-12-03 | 2022-03-01 | 上海交通大学 | Conversation news recommendation method and system based on user implicit feedback |
| JP2023091701A (en) * | 2021-12-20 | 2023-06-30 | ネイバー コーポレーション | Method for place personalization recommendation based on user preference, computer system, and computer program |
| JP7399216B2 (en) | 2021-12-20 | 2023-12-15 | ネイバー コーポレーション | Methods, computer systems, and computer programs for personalized place recommendations based on user preferences |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109189944A (en) | Personalized recommending scenery spot method and system based on user's positive and negative feedback portrait coding | |
| CN110555112B (en) | Interest point recommendation method based on user positive and negative preference learning | |
| CN107679661B (en) | Personalized tour route planning method based on knowledge graph | |
| CN107729444A (en) | Recommend method in a kind of personalized tourist attractions of knowledge based collection of illustrative plates | |
| CN110288436A (en) | A kind of personalized recommending scenery spot method based on the modeling of tourist's preference | |
| CN105320719B (en) | A kind of crowd based on item label and graphics relationship raises website item recommended method | |
| Lin et al. | Heterogeneous knowledge-based attentive neural networks for short-term music recommendations | |
| CN104834679B (en) | A kind of expression of action trail, querying method and device | |
| CN110263256A (en) | Personalized recommendation method based on multi-modal heterogeneous information | |
| CN107657034A (en) | A kind of event social networks proposed algorithm of social information enhancing | |
| CN108804689A (en) | The label recommendation method of the fusion hidden connection relation of user towards answer platform | |
| CN105069717A (en) | Personalized travel route recommendation method based on tourist trust | |
| Hashemi et al. | Overview of the TREC 2016 Contextual Suggestion Track. | |
| CN105718576B (en) | Personalized location recommender system relevant to geographical feature | |
| CN109978660A (en) | A kind of recommender system off-line training method based on intensified learning frame | |
| CN109992674B (en) | Recommendation method fusing automatic encoder and knowledge graph semantic information | |
| El Mohadab et al. | Predicting rank for scientific research papers using supervised learning | |
| CN107016566A (en) | User model construction method based on body | |
| CN107038184A (en) | A kind of news based on layering latent variable model recommends method | |
| CN107391582A (en) | The information recommendation method of user preference similarity is calculated based on context ontology tree | |
| CN109949174A (en) | A kind of isomery social network user entity anchor chain connects recognition methods | |
| CN110083764A (en) | A kind of collaborative filtering cold start-up way to solve the problem | |
| CN108172047A (en) | A kind of network on-line study individualized resource real-time recommendation method | |
| CN107357845A (en) | A kind of tour interest commending system and recommendation method based on Spark | |
| CN106528812A (en) | USDR model based cloud recommendation method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190111 |